# 📜 Competition Rules

Welcome to the **Embodied Agent Interface Challenge @ NeurIPS 2025**!  
To ensure fairness, transparency, and reproducibility, all participants must adhere to the following rules.

## ✅ Eligibility

- Open to **all individuals and teams worldwide** — including students, academics, and industry practitioners.
- No restrictions based on geography or affiliation.
- Multiple teams from the same institution are allowed but must submit **independent entries**.

## 👥 Team Participation

- Participants can join **individually or as part of a team**.
- Each team must designate a **primary contact** for communications.

## 📦 Submission Formats

All interfaces and submission formats are detailed in the [Participate page](Participate).

## 📈 Submission Limits

- A maximum of **10 submissions per team per day** during the **development phase**
- This limit helps avoid leaderboard overfitting and ensures fair resource usage

## 🏁 Final Evaluation Phase

- The **final leaderboard** will be based on performance on a **hidden test set** unavailable to participants.
- No manual tuning, hard-coding, or test set engineering allowed

## 🔁 Reproducibility

To qualify for final ranking and awards:
- Submissions must include all code, dependencies, and setup files
- Top teams’ submissions will be **re-executed by organizers** to verify results
- Models must be fully **automated and deterministic**

## 🌍 Use of External Resources

- You **may use external datasets or models** if:
  - They were publicly available before the competition launch
  - They are clearly cited in your final write-up

## 🚫 Disallowed Practices

Violations will result in **disqualification**. These include:
- Using manually labeled outputs on test sets
- Hard-coding answers or exploiting leaderboard feedback
- Leaking or reverse-engineering the hidden test set
- Submitting outputs that do not generalize to new inputs

## ⚖️ Ethics & Conduct

All participants must follow the **[NeurIPS Code of Ethics](https://neurips.cc/public/EthicsGuidelines)**, including principles of:

- Fairness
- Transparency
- Safety
- Responsible AI development